4 ms·
Tell me - are there any possibilities of advancement in today's streamlined datacenter job? In the past, someone eager and capable might have been given trainin
by HarryHirsch 6y ago
Tell me - are there any possibilities of advancement in today's streamlined datacenter job? In the past, someone eager and capable might have been given training opportunities, but that doesn't really exist any longer, does it?
- throwaway0a5e 6y agoIt's not like they promote people on a schedule but BigCos typically use the lower levels as part of the hiring pool for the higher levels. This has benefits since you know that anyone already working that role already knows your policies and procedures and are basically pre-vetted so you don't have to worry about getting them 90% of the way through the pipeline then failing the drug test or background check. There's basically no chance of moving into a white collar role internally unless you get a degree but there's definitely room for advancement. Though you could probably advance faster by jumping ship.
- caseysoftware 6y agoNope, doesn't look like it. Google is screwing over tons of people here and obviously they're not alone. Why do we "trust" them again?
- username90 6y agoTraining unskilled workers and giving them normal professional worker wages is screwing them over how? It isn't like they would be better off at mcdonalds.
- caseysoftware 6y agoTeaching them skills and laying them off without cause is screwing them over. If you also consider that the skills they teach aren't enough to get them to the next level, it's worse. This is covered in great detail in the article.
- filoleg 6y agoNot sure about Amazon datacenters, but I know an actual person in real life who went from an unskilled Amazon warehouse worker to a full-time SDE at Amazon with zero previous coding experience or a relevant degree. All in a 1-year timespan iirc, simply going through one of the training programs that their warehouse offered. Apparently, those on-the-job training programs that Amazon offers actually tend to work at times. And of course this is just a single datapoint not indicative of anything, but it kinda shows that those programs might actually work to a degree. I don't doubt that their success rate is likely not even close to 100%, but if it worked even for a chunk of the people attempting it, it is still something.